Printer Friendly

Exchange offers and stock swaps - new evidence.

* Exchange offers and swaps are noncash transactions that alter the capital structure of the firm without changing its current portfolio of assets.(1) For this reason, they are thought to be "pure" financial events. There is nothing, of course, that says an exchange offer announcement cannot be correlated with future changes in the firm, e.g. with expected changes in cash flows and investment policy. These future changes are the focus of this study. We find that leverage-increasing exchange offers are associated (i) with decreases in the systematic risk of equity; (ii) with increases in share-adjusted earnings, sales, and total assets; and (iii) with insider purchases of stock prior to the announcement. Opposite effects are found following leverage-decreasing announcements. To the best of our knowledge, these three findings represent new empirical evidence, relevant to the recapitalization puzzle, that has not been previously published elsewhere.

The first section of the paper reviews six theories that have been proposed to explain exchange offers. Section II describes how our sample was chosen; the results of time series tests of the change in systematic risk; insider trading; time series tests fo changes in adjusted earnings, sales and capital expenditures per share; and cross-sectional tests. Section III summarizes and concludes.

I. Possible Explanations for Exchange

Offers and Swaps

Exhibit 1 compares six explanations for exchange offers and swaps. Although these theories are not necessarily mutually exclusive, the best theory is the one that explains the greatest number of phenomena without having any of its predictions rejected. There are four broad categories of predictions: returns of the announcement data, changes in systematic risk, financial data effects, and ownership effects. Exhibit 1 is a useful guide for the theoretical discussion that follows. Predictions that are starred (sup.*) are proven, later in the paper, to be inconsistent with the empirical evidence.

[TABULAR DATA OMITTED]

The naive accounting hypothesis of exchange offers often cites short-term increases in earnings per share as the motivation for exchange offers - whether they are leverage-increasing or leverage-decreasing. While this accounting hypothesis is often viewed as a "straw man" in academia, it is still a valid consideration on Wall Street. To illustrate, we found two newspaper quotes from exchange offers in our sample. The first is leverage-increasing and the second is leverage-decreasing. Not surprisingly, both imply increasing earning sper share in the short run.

NVF Co. said it plans an offer to holders to

exchange new 10% subordinated debentures, due

2003, for up to 500,000 outstanding common

shares ... as a result of the offer shareholders

exchanging their securities may receive debentures

having a high market value ... At the same

time, less stock outstanding would probably

boost per-share earnings. [Wall Street Journal,

October 24, 1973, p. 14;1]

Aluminum Co. of America said it agreed to

swap as many as 1,750,000 new common shares

for ALCOA debentures held by Salomon Brothers

Inc..... The exchange will result in a $22

million first quarter net gain ... the difference

between the deventures' cost and their face value

will account for the $22 million net gain ....

While a nonrecurring gain, the $22 million will

be timely for ALCOA. [Wall Street Journal, January

25, 1982, p. 40;5]

The native accounting hypothesis predicts that an improvement in short-term earnings will elicit a positive response from the stock market whenever earnings per share increases, regardless of the direction of the leverage change. In Section II, we present evidence that, in fact, the opposite is true. The sign of the announcement effect is determined by the direction of the leverage change, not by the change in short-term earnings per share.

The tax-savings hypothesis, proposed by Modigliani and Miller [21], predicts that the interest tax shield provided by leverage-increasing exchange offers (and lost with leverage-decreasing exchange offers) explains the positive (negative) announcement effect. Unfortunately, the signalling hypothesis (explained later) makes exactly the same qualitative directional predictions. Although Pinegar and Lease [23] report positive announcement effects for leverage-increasing exchanges of preferred for common stock - a transaction with no tax shield effects - the relatively smaller magnitude of the announcement effects compared with debt-for-common exchange offers means that one cannot, a priori, rule out the possibility that debt-for-common exchange offers have both tax and signalling effects.

Since one of the important empirical issues is how beta might change given the announcement of an exchange offer, we want to study the implications of the tax-savings hypothesis in this regard. Equation (1) is the "textbook" relationship between the levered beta, "[beta.sub.L]," and the unlevered beta, "[beta.sub.upsilon]",

[beta.sub.L] = [1 + (1 - T)(B/S)][beta.sub.upsilon]

in a world with corporate taxes as implied by Hamada [11]. It predicts, ceteris paribus, that greater leverage, as measured by the ratio of the market value of debt to the market value of equity, B/S, will result in higher risk for equity holders, i.e., "[beta.sub.L]" will increase. The usual ceteris paribus assumption is that neither the marginal statutory tax rate, T, nor the market value of equity, S, change. However, even with the maximum theoretical increase in shareholders' wealth predicted by the Modigliani and Miller proposition [21] the levered beta will increase.(2) Therefore, the tax-savings hypothesis unambiguously predicts that beta will increase with leverage-increasing exchange offers. (3)

The debtholder wealth expropriation hypothesis proposes that shareholders encourage leverage-increasing exchange offers to expropriate debtholders' wealth (ex post) by exposing them to greater risk. However, when adequate, protective covenants can prevent expropriation both ex ante and ex post. Of course, this is not the case in every instance, nevertheless Dann [6] and Vermaelen [27] find no significant bondholder wealth losses when leverage-increasing stock repurchases are announced. Evidence in direct contradiction to the bondholder expropriation hypothesis is reported by Cornett and Travlos [5] who find statistically significant negative announcement returns for straight debt during leverage-decreasing exchange offers. Furthermore, it is hard to understand why shareholders would support leverage-decreasing offers that would result in expropriating their own wealth for the benefit of bondholders.

The free cash flow hypothesis, popularized by Jensen [14], proposes that debt creation without retention of the proceeds of the issue (as in an exchange offer or swap) enables managers to effectively bond their promise to pay out future cash flows rather than investing them in negative net present value projects or acquisitions. By issuing debt in exchange for stock, managers are bonding their promise to pay out future cash flows in a way that cannot be accomplished by simple dividend increases. In doing so, they give shareholder recipients of the debt the right to take the firm into bankruptcy court if promised interest and principal payments are not made. The free cash flow hypothesis predicts positive announcement date equity returns for leverage-increasing events because a useful constraint is placed on management, and negative shareholder returns are predicted for leverage-decreasing events because the constraint is being relaxed.

One must be careful to stipulate that the free cash flow hypothesis will not be relevant for firms with the need to use cash flows to invest in positive net present value projects. Given this qualification, the free cash flow hypothesis predicts that capital expenditures decrease following the announcement of the transaction. Another inference is that leverage-increasing firms will have lower insider ownership because it is the separation of ownership and control that gives rise to the need for bonding. If the firm were 100% owned by a single individual, there would be no need for bonding. Finally, a limitation of the free cash flow hypothesis is that it does not provide an explanation for why shareholders would support leverage-decreasing events (that are, presumably, anticipated to decrease shareholders' wealth).

The dynamic capital structure hypothesis proposed by Fischer, Heinkel, and Zechner [8] defines an optimal capital structure range that results from the tradeoff between the tax benefits of leverage and bankruptcy costs, given a dynamic setting where asset values change continuously, and where recapitalization is costly. Their model implies that smaller, riskier, lower-tax, lower bankruptcy cost firms will exhibit wider swings in their debt ratios over time. Their empirical evidence is consistent with their predictions. However, when strictly interpreted, their model predicts zero announcement effects because all information is continuously and publicly available, and of course this is inconsistent with the empirical evidence of the aforementioned event studies. If one were to extend their model to assume that leveraged recapitalizations occur with a lag, then the optimal boundary would be exceeded, and presumably there would be an announcement effect as the firm moves from suboptimal to optimal leverage. The predicted announcement effect would be positive for leverage-decreasing recapitalization - and therefore inconsistent with the empirical evidence. Alternately, if one were to modify their model with asset values that change rapidly (in a jump) and to assume that firms simultaneously recapitalize, then leverage increases (decreases) would be associated with equity return increases (decreases) - a prediction that is consistent with the evidence. And finally, if asset value jumps are followed by recapitalizations, then regardless of the direction that asset values change, recapitalization returns the firm to its optimal range and, we would expect leverage-decreasing exchange offers to be associated with positive returns, a prediction that is inconsistent with the facts.

Signalling hypotheses are in our opinion, the most promising for explaining exchange offers. Our model is in the spirit of the incentive-compatible models of signalling equilibrium developed by Ross [25], Leland and Pyle [17], and Heinkel [13].[4] We assume that insiders have superior information regarding the future prospects of the firm, and that they do not participate in leverage-increasing exchange offers when debt is used to repurchase equity from outside (relatively less informed) shareholders.[5] Hence, inside ownership becomes more concentrated as a consequences of the recapitalization.(6) Given information asymmetry, and the additional assumption that the exogenously imposed cost of greater financial leverage is borne ex ante by insiders, we can show that exchange offers will be interpreted as costly signals of the future prospects of the firm. A signalling equilibrium results. Firms with good prospects will signal with leverage increases and those with bad news will decrease leverage to avoid the greater penalty imposed by bankruptcy.

Once the signal is revealed, the market can react by increasing the value of leverage-increasing firms. One of the interesting implications is that they systematic risk of equity may decrease following a leverage-increasing exchange offer.(7). Although the levered equity beta must increase when the face value, F, of debt outstanding increases, it decreases with the market value of the firm, V, and with the maturity of debt, Time,

[Mathematical Expression Omitted]

The signalling model allows for decreases in beta following leverage-increasing exchange offers and vice versa. Later on, our empirical results show that this is actually what happens.

The signalling model also predicts that insiders can benefit from the asymmetric information in their possession. Hence, we should expect net insider purchases prior to leverage-increasing events, and net insider sales before leverage decreases. The other theories (implicitly) assume symmetric information, and therefore make no prediction regarding insider trading behavior.

If recapitalizations are signals about the future prospects of the firm we would expect that both the expected and unexpected components of earnings should go up following leverage-increasing events (in spite of higher interest expenses). In the empirical work that follows, we also examine changes in sales and in capital expenditures, although the signalling theory makes no predictions in this regard. For example, capital expenditures may go down relative to prior expenditures because the company may have made a large expenditure last year and learned this year that operating results were better than expected due to the new investment. If so, we would observe that leverage-increasing exchange offers are followed by lower capital expenditures. The opposite, however, may also be true.

II. Empirical Evidence

A. Data Selection

Our sample consists of all exchange offers and swaps for New York or American Stock Exchange listed firms for the 22-year period from mid-1962 to mid-1984 that met the following criteria:

- The initial announcement data was clearly identified

in the Wall Street Journal Index.

- The offer or swap altered the level of debt or

preferred stock outstanding.[8]

- Cash and other asset distributions associated

with the offer were limited to a maximum of

25% of the value of the affected securities in the

offer.

- No other announcement within one week on

either side of the exchange offer was permitted.

- There was complete return data for the 530

trading days surrounding the announcement.

Exhibit 2 summarizes the transactions included in our sample.[9] There is no evidence that exchange offers or swaps were clustered in time.
Exhibit 2. Exchange Offers and Swaps Between Mid-1962
 and Mid-1984, Categorized by Type
 Leverage-Increasing
Debt-for-common 49
Debt-for-preferred 27
Preferred-for-common 14
 __
Total 90
 Leverage-Decreasing
Common-for-debt 88
Preferred-for-debt 11
Common-for-preferred 28
 ___
Total 127


Daily rates of return were obtained from the CRSP daily return tape for 265 days prior to and 265 days following the initial announcement of the event. Both the CRSP equally weighted and value-weighted market return indices were used.

B. Time Series Tests of Changes in Systematic

Risk

We tested for changes in beta in two steps. First, we found evidence of nonstationarity around the event, and then we employed a dummy-variable approach to determine the direction and significance of changes in the intercept and slope of a market model regression.

The Chow [4] F-test can be used to determine non-stationary. The sum of squared residuals (SSR) from the market model regression, [R.sub.it] = [a.sub.i] + [b.sub.i.R.mt] + [e.sub.it], (3) are estimated from daily return data running from 265 days up to 15 days before the announcement. The SSR are then estimated separately for the interval + 15 to + 265 days after the announcement, and for the pooled returns. An F-test with (250,248) degrees of freedom,

[Mathematical Expression Omitted]

is significant at the 5% confidence level if greater than 1.25. For the sample of 90 leverage-increasing events, the F-test was significant 34 times, whereas at the 5% confidence level one would expect it to be significant only 4.5 times. For the sample of 127 leverage-decreasing events, it was significant 56 times versus 6.35 expected. Thus, the Chow test indicates significant nonstationarity in the market model relationship.[10]

Given evidence of nonstationarity, we tested for changes in the slope and intercept of the market model by using the following dummy variable regression for each stock in our sample,

[R.sub.it] = [a.sub.1i] + [a.sub.2i.D.sub.it] + [b.sub.1i.R.sub.mt] + [b.sub.2i.D.sub.it.R.sub.mt] + [e.sub.it.] (4)

The dummy variable, [D.sub.it], takes the value zero prior to the announcement (t = -265 to -15) and the value one after the event (t = +15 to +265). A standard t-test on the dummy variables is used to determine the significance of changes in the intercept and slope. The results are summarized in Exhibit 3. In 61 out of 90 leverage-increasing transactions the beta went down, and in 75 of 127 leverage-decreasing transactions it went up! The direction of these beta changes is exactly the opposite from what traditional articles that incorporate tax savings predict (e.g., Hamada [10, 11] and Galai and Masulis [9]). It is consistent, however, with the proposition that leverage-increasing exchange offers are signals of improving prospects of the firm.

[TABULAR DATA OMITTED]

Healy and Palepu [12] find results for a large sample of primary stock offerings that are very similar to ours. Following these leverage-decreasing events, they report statistically significant increases in beta. They suggest that "One interpretation of these findings is that managers decide to issue equity and reduce financial leverage when they foresee an unexpected increase in their firm's business risk."

A binomial z-statistic under the null hypothesis that there is a fifty-fifty chance of increases in the slope and intercept,

[Mathematical Expression Omitted]

was used to test for significance in our sample. The z-statistics for changes in beta for both the leverage-increasing and the leverage-decreasing transactions were significant at the 5% level. The change in intercept was significant only for the leverage-decreasing events. When one counts the direction of change, however, only 8 of 127 changes were significant versus an expected 6.35 at the 5% confidence level.

Exhibit 4 shows summary statistics broken down by type of transaction. Column 3 is the average of the changes in beta estimated from the dummy regression. Columns 4 and 5 give the OLS market model regression betas for separate regressions before and after the transaction. Columns 6 and 7 show Scholes-Williams [26] betas to correct for bias in the betas that may be caused by nonsynchroniety in the daily data. All three procedures produce similar results regarding the direction and magnitude of changes in beta. Note that although leverage-decreasing firms (row 8) were roughly ten times the size of leverage-increasing firms (row 3) the observed change in beta was nearly identical. This leads us to conclude that the smaller size of leverage-increasing firms did not substantially bias our estimates of beta.

[TABULAR DATA OMITTED]

A Wilcoxon nonparametric rank sum test was used to compare the distribution of pre-event with post-event betas and to compare the OLS betas with the Scholes-Williams betas. The test statistic, the Wilcoxon T, has an asymptotically unit normal distribution. Exhibit 5 summarizes the results which support the conclusion that there is no significant difference between the OLS and Scholes-Williams betas, and that either measure of beta indicates lower systematic risk following leverage increasing events and vice versa.

Exhibit 4 also gives the two-day abnormal returns for the announcement period, estimated as the difference between the actual returns and those predicted from a market model regression estimated during the pre-event benchmark period.[11] Two-day returns were used because the actual news release may have occurred while the market was open on the announcement date (day 0) or after the close of the market so that the return would be reflected the next day (day 1). The largest positive two-day returns, 8.0%, were for debt-for-common transactions and the largest negative returns, -6.4%, occurred with common-for-debt exchange offers. This result is the same order of magnitude as other studies, e.g., 0.79% reported by Masulis [19]/ In general, the largest announcement data returns are followed by the largest changes in beta. It is interesting to compare the change in size of the offer (6% change in book value leverage) with the size of the announcement effect (8% of equity) for debt-for-common transactions. The dollar size of the announcement effect is large than the total size of the transaction. The theoretical maximum tax savings at any time during our

(1) Stock-for-debt exchange offers were replaced by stock-for-debt swaps after July 1981 to avoid a tax liability that would arise if the firm were to directly repurchase its outstanding debt at a discount. In a stock-for-debt swap, an investment bank would purchase the firm's bonds, accept stock as payment, then resell the stock. This form of swap was classified as a tax-free recapitalization. In 1984, the Tax Reform Act eliminated even this tax-favored treatment and swaps have been used infrequently since then. See Finnerty [7] for further details. (2) Using the assumptions of Modigliani and Miller [21], proof is straightforward. First, the value of the levered firm, [V.sub.L], is equal to the value of the unlevered firm, [V.sub.u], plus a gain from leverage, TB.

[V.sub.L] = [V.sub.u] + TB S + B = [V.sub.v] + TB

dS = -(1 - T)dB

Nest, the derivative of Equation (1) with respect to the level of debt, B, is

d([beta.sub.L]/dB = (1 - T)[beta.sub.u][d(B/S)/dB]
 = (1 - T)[beta.sub.u][S - B(dS/dB)]/[S.sub.2]
 = [(1 - T)[beta.sub.u]/S] {1 + [B(1 - T)/S]}
 = [(1 - T)[beta.sub.u]/S] {[S + B(1 - T)]/S}
 = [(1 - T)[beta.sub.]/S] [V.sub.u]/S] > 0


Therefore, the tax-savings hypothesis unambiguously predicts that the levered beta will increase, given an increase in leverage. (3) If we introduce personal taxes, the value of the levered firm is the same as the value of the unlevered firm in a Miller equilibrium. Hence, in footnote 2, d(B/S)/dB is positive and so too is the change in the levered beta. (4) The formal model, not presented here in order to conserve space, is available from the authors. (5) This fact was pointed out by Masulis [19, p.308]. (6) Vermaelen [27, p. 178] and [28] emphasized that (leverage-increasing) exchange offers are very similar in their financial impact to share repurchases via tender offers that are financed by issuing debt. (7) For proof, see Galai and Masulis [9]. (8) For example, offers whose sole effect was to change the terms of debt outstanding (coupon rates, covenants, or maturity) were excluded from the sample. (9) Masulis [20] collected 133 exchange offers that meet the same screening criteria during the mid-162 to mid-1978 time interval. Our sample differs from his primarily by adding 75 stock-for-debt swaps that occurred between mid-1981 and mid-1984 and by adding 15 exchange offers that occurred between 1970 and 1984. (10) Similar results were obtained using a Lagrange multiplier test. See Breusch and Pagan [1] for an explanation of the methodology. (11) the average abnormal return estimates were not significantly different when estimated using Scholes-Williams betas, or when using a post-event benchmark period.
Exhibit 5. Wilcoxon Rank Sum Tests for the Leverage-Increasing
 Increasing and Leverage-Decreasing
 [Samples.sup.a]
 Wilcoxon T-Statistic
 Leverage- Leverage-
 Hypothesis Increasing Decreasing
Pre- and post-event OLS
 betas are equal [2.09.sup.b] [-1.78.sup.b]
Pre-and post-event
 Scholes-Williams
 betas are equal 1.56 [-2.04.sup.b]
OLS and Scholes-Williams
 betas are equal pre-event -0.15 1.25
OLS and Scholes-Williams
 betas are equal post-event -0.43 0.68
(a) There are 90 observations in the leverage-increasing sample and 127
in the leverage-decreasing sample.
(b) Statistically significant at the 10% confidence level (T=1.65).


sample period could be only as much as roughly 50% of the exchange offer. The fact that the increase in the value of equity exceeds the maximum that could be attributed to tax savings is strongly consistent with signalling.

Another interesting fact is that leverage-increasing firms were about one-tenth the size of leverage-decreasing firms. Later on, we shall see that they also have growing earnings while leverage-decreasing firms have decreasing earnings, and that the percentage of insider ownership is significantly higher. These facts are consistent with the interpretation that exchange offers are really noncash share repurchases designed to increase the concentration of ownership in small firms with good future prospects (the signalling hypothesis), rather than a bonding mechanism to reduce agency costs between management and shareholders (the free cash flow hypothesis).

C. Insider Trading Prior to the Announcement

If the signalling hypothesis is true, one might expect, SEC regulations notwithstanding, that there might be net purchases by insiders prior to the announcement of leverage-increasing exchange events and vice versa for leverage-decreasing events.(12) The insider decisions to buy or sell holdings are compiled from the Value Line Insider Decision Index as reported in the Value Line Investment Survey since 1972. The insider decision index measures the net buy and sell decisions of insiders during the previous two quarters.

Exhibit 6 shows the results for 126 transactions that occurred after 1972. In 36 out of 40 leverage-increasing events there were net insider purchases prior to the announcement. In addition, the changes in beta were significantly negative and the announcement data returns were significantly positive. Prior to leverage-decreasing events, there were net insider sales in 56 of 86 cases, the change in beta was significantly positive, and the announcement date returns were negative, but not significant.(13)
Exhibit 6. Insider Transactions for the Two Quarters
 Preceding Exchange Offers and Swaps
 Net Insider Action
Type of Transaction Purchases Sales
Leverage-Increasing: 36 4
 Change in beta -0.2556 -0.0961
 t-statistic (-2.70) (-0.44)
 Two-day AR 0.0455 0.0942
 t-statistic (3.90) (0.96)
Leverage-Decreasing: 30 56
 Change in beta 0.1451 0.1444
 t-statistic (2.10) (2.32)
 Two-day AR -0.0022 -0.0068
 t-statistic (-0.39) (-1.43)


The fact that insiders buy prior to leverage-increasing exchange offers and swaps and sell prior to leverage-decreasing events is consistent with the signalling hypothesis, and is not necessarily inconsistent with other hypotheses, except for the fact that they all implicitly assume symmetric information. Of course, we cannot rule out that events during the two quarters prior to the leverage change and correlated with it were actually responsible for insider behavior. For example, insiders who believe the value of common stock will increase due to the tax shield provided by leverage-increasing events will rationally increase their share ownership, if they also believe that they possess private information.

Next, we turn to time series tests of changes in earnings, sales, and capital expenditures. If these show unexpected improvement then the market is correct when it interprets leverage-increasing exchange offers as good news.

D. Time Series Tests of Earnings, Sales, and

Capital Expenditures

Value Line forecasts of earnings per share have been shown to be superior to the best time series models.(14) Using the Value Line Investment Survey, we collected forecasts one year ahead for earnings, capital expenditures, sales, and the book value of total assets per share; as well as their realizations. As shown in Exhibit 7, about two-thirds of the original sample of 217 were located. Because Value Line forecasts did not anticipate exchange offers or swaps, it was necessary to adjust their figures for the change in shares outstanding. For example, unexpected earnings per share were defined as: [eps.sub.u] = [eps.sub.a] - [eps.sub.fadj] (6) [eps.sub.fadj] = [eps.sub.f] x ([n.sub.p]/[n.sub.o]) (7) where [eps.sub.u] = unexpected eps [eps.sub.a] = actual post-event eps [eps.sub.f] = forecast of eps [eps.sub.p] = pre-event eps [eps.sub.fadj] = adjusted forecast eps [eps.sub.padj] = adjusted pre-event eps [n.sub.p] = pre-event number of shares [n.sub.o] = post-event number of shares.
Exhibit 7. Sample Size of Value Line Forecast Data by
 Transaction Type, Mid-1962 to Mid-1984
 Original Value Line
Leverage-Increasing:
 Debt-for-common 49 30
 Debt-for-preferred 27 20
 Preferred-for-common 14 9
 Total 90 59
Leverage-Decreasing:
 Common-for-debt exchange offer 13 5
 Stock swaps 75 70
 Preferred-for-debt 11 6
 Common-for-preferred 28 21
 Total 127 102


Similar adjustments were made for sales, capital expenditures and total assets per share so that all results are reported on a fully diluted basis. Also, expected changes in earnings per share were calculated as: [DELTA] eps =[eps.sub.fadj] - [eps.sub.padj] (8) [eps.sub.padj] = [eps.sub.p] x ([n.sub.p)]/[(n.sub.o]) (9)

Results are presented in Exhibit 8.(15) The leverage-increasing and leverage-decreasing samples are remarkably different, except that both have roughly the same effective tax rate - a fact that somewhat weakens support for the tax hypothesis because leverage-decreasing firms seem to be sacrificing an interest tax shield, at least during the year of the exchange offer.

[TABULAR DATA OMITTED]

There are a number of facts in Exhibit 8 that are consistent with the signalling hypothesis. Insider ownership is much greater for leverage-increasing firms (20.6% versus 8.6%). Expected changes in earnings, capital expenditures, sales, and total assets per share are all positive for leverage-increasing firms and unexpected changes in earnings and sales per share are also positive. Unexpected changes in capital expenditures are negative, but as we discussed earlier, this is not necessarily inconsistent with signalling. On the other hand, the expected changes in earnings and capital expenditures per share for leverage-decreasing firms are negative. Unexpected changes in all four variables are negative with the unexpected changes in earnings per share being statistically significant at the 5% confidence level (t=-4.67). Furthermore, the expected changes in sales and total assets per share, although positive, as much less than for leverage-increasing firms.

Except for the fact that the unexpected change in capital expenditures is significantly negative for leverage-increasing events, this evidence seems largely inconsistent with the free cash flow hypothesis. Firms that undertake laverage-increasing exchange offers and swaps are much smaller, more closely held by a significant margin, and seem to have bright futures. they seem to have less reason for a conflict between management and shareholders than do leverage-decreasing firms. the transaction does not appear to be intended to bond management to its promise to pay out future cash flows. there is little evidence to support the supposition that firms undertaking leverage-increasing exchange offers do so in order to keep from using cash to undertake negative NPV projects. In particular, the unexpected increase in earnings per share is more consistent with the signalling hypothesis and that leverage-increasing exchange offers are best iterpreted as maneuvers to use debt (rather than cash) to repurchase shares and further concentrate inside ownership.

E. Cross-Sectional Explanations of Abnormal

Returns and Changes in Beta

To compare various explanations of announcement date abnormal returns and to study the partial effects of independent variables, the following cross-sectional regression was run on various partitions of the data,

[AR.sub.i] = [a.sub.1] + [a.sub.2][D.sub.i]+[b.sub.1](% change in [eps.sub.i]) + [b.sub.2](% [insider.sub.i) + [b.sub.3]([tax.sub.i]) + [b.sub.4]([maturity.sub.i]) + [b.sub.5](face[value.sub.i]) + [e.sub.i]. (10) Similar regressions were run with the change in beta as the dependent variable. The dummy variable, [D.sub.i], measures the mean effect of a given data partition while holding other variables constant, e. g., the size of the offer. For example, the signalling hypothesis is tested in the first row of Exhibit 9. The dummy is equal to one if the transaction was leverage-increasing and equal to zero otherwise. It is positive and statistically significant, indication that leverage-increasing exchange offers and swaps are signals resulting in positive announcement returns. The second independent variable, the percent change in earnings per share (% change in eps), was measured three ways: (i) as the actual total percentage change in eps, (ii) as the expected percentage change in adjusted eps, and (iii) as the unexpected percentage change in adjusted eps. None of these variables are significant in the first three rows of Exhibit 9, indicating that the marginal effect of eps on the signal is immaterial. The percent of shares held by insiders, % insider{sub.i], is introduced as an independent variable in as attempt to measure one aspect of the free cash flow hypothesis, namely that the greatest bonding effect for leverage-increasing exchange offers would occur when insider ownership is low. None of the insider coefficients is significant. Therefore, this variable lends no support to the free cash flow hypothesis. The fourth independent variable, [tax.sub.i], measures the potential tax savings by multiplying the firm's effective tax rate by the size of the exchange offer.(16) Measured on a per-share (fully diluted) basis, it was statistically insignificant in all regressions and lends no support to the tax hypothesis. The fifth variable, [maturity.sub.i], is the difference between the pre- and post-event book-value weighted average number of years to maturity, i.e., the change in maturity.(17) One might expect that lengthening the maturity structure of debt could benefit shareholders. The sixth and last variable, [face.sub.i], the the dollar amount of the change in the face value of debt or preferred in the transaction.

[TABULAR DATA OMITTED]

The signalling hypothesis predicts that leverage-increasing transactions will be associated with positive announcement date returns and with decreased betas. The first three rows of Exhibit 9 and 10 test these predictions. The dummy variable in Exhibit 9 is positive and significant, indicating the predicted relationship between announcement date returns and the leverage change of the transaction. In exhibit 10, the negative significant dummy variable is consistent with beta decreasing when leverage increases - a result consistent with signalling. Note also that Exhibit 11 shows that in almost every partition of the data, decreasing betas are also associated with higher earnings following the transaction.

[TABULAR DATA OMITTED]
Exhibit 11. Announcement Date Returns and Changes
 in Beta Partitioned by the Sign of the
 Leverage and the eps Change
 Fully Diluted Change in eps (total)
 Positive Negative
Leverage-Increasing 51 8
 Announcement return 0.0377 (4.29) 0.041 (1.75)
 Change in beta -0.2603 (-2.93) -0.1970 (-1.80)
Leverage-Decreasing 32 70
 Announcement return -0.0118 (-2.08) -0.0041 (-0.92)
 Change in beta 0.0367 (0.47) 0.1167 (1.80)


The naive accounting hypothesis predicts that all exchange offers should increase short-term earnings per share, or at least that if earnings per share increases as a result of the exchange offer, the announcement effect should be positive. Exhibit 11 shows a partition of announcement returns and changes in beta by the direction of leverage change (the signal) and by the direction of the change in eps. When the change in eps is positive and leverage increases, the announcement return is positive and significant, and beta decreases. The results for decreasing eps and leverage are also as expected. However, when there is a conflict between the signalling effect and the one-year eps change, the signalling effect dominates.(18) For example, when the one-year change in eps is positive but leverage decreases (an unfavorable signal about all future cash flows), the announcement return is significantly negative and beta increases. Thus the evidence is consistent with signalling, but not with the naive accounting hypothesis. Exhibits 9 and 10 provide confirmation in the form of multiple regressions. Rows 4 through 6 of Exhibit 9 show that when the sample is partitioned on the basis of the eps change, the dummy is marginally significant at the 10% level in one case (total eps) but has the wrong sign. In Exhibit 10, the dummy (in rows 4-6) is insignificant.

The tax benefit hypothesis predicts that firms can increase their tax shelters by increasing leverage. To test the tax hypothesis we separated transactions that had debt involved, and therefore a potential tax shelter, from those that had no debt involved (e.g., a preferred-for-stock exchange offer). The results are given in rows 7-9 of Exhibits 9 and 10 for leverage-decreasing transactions and 10-12 for leverage-decreasing transactions. The dummy variable is insignificant for the leverage-increasing sample and significant with the wrong sign for the leverage-decreasing sample. The results lend no support for the tax hypothesis, and even contradict it for leverage-decreasing events.

The debtholder wealth expropriation hypothesis predicts that leverage-increasing transactions may benefit shareholders by diluting the position of debtholders who in many cases have imperfect protective covenants. To test the hypothesis, we partitioned the sample into senior-for-junior security transactions, e.g., debt or preferred for stock, and senior-for-senior transactions, e.g., debt-for-preferred. None of the dummy variables in Exhibits 9 or 10 are significant, either for leverage-increasing transactions (rows 13-15) or for leverage-decreasing transactions (rows 16-18). The marginally significant results for leverage-decreasing transactions (Exhibit 9, rows 16-18) have the wrong sign. We conclude, therefore, that there is little support in the data for the wealth expropriation hypothesis.

III. Summary and Conclusions

The empirical evidence presented in this paper leads to the conclusion that leverage-increasing exchange offers and swaps may be interpreted as repurchases of equity by issuing debt. Insiders, who do not participate in the repurchase, use their superior information regarding the future prospects of the firm to increase their ownership position. They don't have the actively buy or sell shares (although our insider trading evidence indicates they do). They merely issue new debt and use the proceeds to repurchase the shares of outsiders. The market is not fooled by the maneuver, and correctly interprets the announcement of the transaction as good news, thereby increasing the wealth of all shareholders at the announcement date, before the transaction is consummated.

The empirical evidence is strongly consistent with signalling as the best interpretation of the data. Leverage-increasing exchange offers have positive announcement effects, are preceded by net insider purchases, and are followed by decreases in systematic risk, increases in (fully diluted) earnings, sales, and total assets per share. Leverage-decreasing exchange offers have the opposite effects.

None of the alternative explanations for exchange offers and swaps are entirely consistent with the empirical evidence. The naive accounting hypothesis is inconsistent with the fact that firms with higher earnings following leverage-decreasing exchange offers have statistically significant negative announcement returns accompanied by increases in beta. The tax hypothesis is weak because it cannot explain several findings. First, it predicts increases in beta following leverage-increasing events, but exactly the opposite actually happens. Furthermore, the tax hypothesis fails to predict positive announcement returns observed in preferred-for-common exchange offers, and it does not predict higher earnings, sales, and total assets per share following leverage-increasing events. It is difficult to reject the tax hypothesis altogether because its effects may be masked by signalling. For example, the announcement date returns of debt-for-common events are larger than the tax hypothesis can explain, but this result may be attributed to the combined positive effects of tax savings and signalling. Also, the observed decrease in beta following leverage-increasing events may result from a large beta decrease attributable to signalling and a small beta increase due to the tax effect. Finally, although the cross-sectional regressions have low power, they do reject the accounting, the tax-benefit, and the bondholder wealth expropriation hypotheses because whenever the dummy variable are significant at standard levels (or marginally significant) they have the wrong sign. It seems that the signalling hypothesis is most consistent with all of the empirical results.

It is prudent, however, to conclude by saying that no single explanation can cover all circumstances and all periods in history. Our sample ((1962-1984) largely excludes events affected by the recent leverage buyout phenomenon. Many of these LBO transactions may have been motivated by superior management information (the signalling hypothesis), but others may have strong bonding incentives (the free cash flow hypothesis), or strong tax incentives (the tax hypothesis). In several situations, bondholders' wealth was expropriated due to inadequate protective covenants.

(12) For a similar use of insider trading to interpret new security issues, see Karpoff and Lee[15]. (13) For leverage-increasing events the binomial z-statistic was 5.06, rejecting, at the 5% confidence level, the null hypothesis that the net purchases of stock by insiders was random with a 50-50 probability. For leverage-decreasing events the z-statistic was 2.80, also rejecting the null hypothesis at the 5% confidence level. (14) For example see Brown, Griffin, Hagerman, and Zmijewski [2 and 3]. (15) Pre- and post-event figures are those actually reported by the company. Expected and unexpected figures are adjusted as shown in Equations (6) - (8). Earnings reflect actual interest expense, and no adjustment was made for nonrecurring gains or losses. (16) The effective tax rate was defined as the cash taxes divided by taxable income. (17) For the stoc-for-preferred and preferred-for-stock exchange offers, the maturity of preferred stock was assumed to be 30 years. (18) Similar results were obtained when the expected change in eps or unexpected change in eps were used instead of the total change. Earnings per share was not adjusted for nonrecurring gains or losses, and reflects actual interest expense.

References

[1.] T. Breusch and J. Pagan, "The Lagrange Multiplier Test and its

Applications to Model Specification in Econometrics," Review

of Economic Studies (January 1980), pp. 239-253. [2.] L. Brown, P. Griffin, R. Hagerman, and M. Zmijewski, "Security

Analysts' Superiority Relative to Univariate Time Series Models

in Forecasting Quarterly Earnings," Journal of Accounting and

Economics (April 1987), pp. 61-88. [3.] L. Brown, P. Griffin, R. Hagerman, and M. Zmijewski, "An

Evaluation of Alternative Proxies for the Market's Assessment

of Unexpected Earnings," Journal of Accounting and Economics

(July 1987), pp. 159-194. [4.] Chow, G., "Tests for the Equality Between Sets of Coefficients

in Two Linear Regeressions," Econometrica (July 1960), pp.

591-605. [5.] M. Cornett and N. Travlos, "Information Effects with Debt-for-Equity

and Equity-for-Debt Exchange Offers," Journal of

Finance (June 1989), pp. 451-468. [6.] L. Dann, "Common Stock Repurchases: An Analysis of Returns

to Bondholders and Stockholders," Journal of Financial Economics

(June 1981), pp. 113-138. [7.] J. Finnerty, "Stock-for-Debt Swaps and Shareholder Returns,"

Financial Management (Autumn 1985), pp. 5-17. [8.] E. Fischer, R. Heinkel, and J. Zechner, "Dynamic Capital Structure

Choice: Theory and Tests," Journal of Finance (March

1989), pp. 19-40. [9.] D. Galai and R. Masulis, "The Option Model and the Risk Factor

of Stock," Journal of Financial Economics (January-March

1976), pp. 53-81. [10.] Hamada, R., "Portfolio Analysis Market Equilibrium and Corporate

Finance," Journal of Finance (March 1969), pp. 13-31. [11.] Hamada, R, "The Effects of the Firm's Capital Structure on the

Systematic Risk of Common Stocks," Journal of Finance (May

1972), pp. 435-452. [12.] P. Healy and K. Palepu, "Earnings and Risk Changes Surrounding

Primary Stock Offerings," Journal of Accounting Research

(Spring 1990), pp. 25-48. [13.] R. Heinkel, "A Theory of Capital Structure Relevance Under

Imperfect Information," Journal of Finance (December 1982),

pp. 1141-1150. [14.] M. Jensen, "Agency Costs of Free Cash Flow, Corporate

Finance, and Takeovers," American Economic Review (May

1986), pp. 323-329. [15.] J. Karpoff and D. Lee, "Insider Trading Before New Issue

Announcements," Financial Management (Spring 1991), pp. 18-26. [16.] W. H. Lee, "The Effects of Exchange Offers and Stock Swaps on

Equity Risk and Shareholders' Wealth: A Signalling Model Approach,"

Unpublished Ph.D. Thesis, University of California,

Los Angeles, 1987. [17.] H. Leland and D. Pyle, "Information Asymmetries, Financial

Structure and Financial International," Journal of Finance

(May 1987), pp. 371-387. [18.] J. McConnel and G. Schlarbaum, "Evidence on the Impact of

Exchange Offers on Security Prices: The Case of Income

Bonds," Journal of Business (January 1981), pp. 65-85. [19.] R. Masulis, "The Effects of Capital Structure Change on

Security Prices: A Study of Exchange Offers," Journal of Financial

Economics (June 1980), pp. 139-178. [20.] R. Masulis, "The Impact of Capital Structure Change on Firm

Value: Some Estimates," Journal of Finance (March 1983), pp.

107-126. [21.] F. Modigliani and M. Miller, "Corporate Income Taxes, and the

Cost of Capital: A Correction," American Economic Review

(June 1963), pp. 433-443. [22.] J. Peavy and J. Scott, "A Closer Look at Stock-for-Debt Swaps,"

Financial Analysis Journal (May-June 1985), pp. 44-50. [23.] J. Pinegar and R. Lease, "The Impact of Preferred-for-Common

Exchange Offers on Firm Value," Journal of Finance (September

(1986), pp. 795-814. [24.] R. Rogers and J. Owers, "Equity for Debt Exchanges and Stockholder

Wealth," Financial Management (Autumn 1985), pp.

18-26. [25.] S. Ross, "The Determination of Financial Structure: The Incentive-Signalling

Approach," Bell Journal of Economics (Spring

1977), pp. 23-40. [26.] M. Scholes and J. Williams, "Estimating Betas for Non-Synchronous

Data," Journal of Financial Economics (December

1977), pp. 309-328. [27.] T. Vermaelen, "Common Stock Repurchases and Market Signalling:

An Empirical Study," Journal of Financial Economics

(June 1981), pp. 139-183. [28.] T. Vermaelen, "Repurchase Tender Offers, Signalling, and

Managerial Incentives," Journal of Financial and Quantitative

Analysis (June 1984), pp. 163-182.
COPYRIGHT 1991 Financial Management Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Topics in Capital Restructuring
Author:Copeland, Thomas E.; Won Heum Lee
Publication:Financial Management
Date:Sep 22, 1991
Words:7190
Previous Article:Investment options, assets in place, and the risk of stocks.
Next Article:The role of private and public debt in corporate capital structures.
Topics:


Related Articles
Executive equity swaps and corporate insider holdings.
Daiei to cut capital by 99% under restructuring plan+.
Daiei's 3 main banks to provide 520 bil. yen in aid+.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters